Bipedal walker

Actual human joints, being non-ideal, exhibit motion in all 6 DOF. ie they have prismatic as well as rotational components. I guess it's just due to the cost vs accuracy law that most designs settle on less.

I'm am actually building a 12 DOF biped for my thesis using a 'stacked gimbal' hip joint. It seems that 6 DOF per leg is the minimum required for reasonable bipedal motion in 3 dimensions...

-Duncan

Mike,

Asimo looks impressive in TV commercials but is as far from real useful bipedal walking as you can possibly get. It seems that all Japanese bipedal walkers use the same technique (which I've named "constipated walking". Knees bent, carefully shift the CG from one leg to the other, then take a step, stop, center the cg, take a dump, switch legs and do it again. Impressive technology integration to be sure, but I suspect it represents a bunch of steps (no pun intended) in the wrong direction.

I think that the guys at MIT are doing it right, but it's going to take a while before they can shoot cheap TV commercials. I visited the famous Leg Laboratory a couple of years back and got to see some of the things they were working on at the time.

But, what do I know. :-)

RU sure? Try dangling your leg over a chair and hold your thigh rock-steady. You'll find yourself capable of side-to-side articulation of the knee.

Perhaps those of us in the field of robotics are all freaks? Or at least distanced from the "Earth People" - we analyse our own movement (carefully avoiding the word, "motions"!!) and also our own brain processes in order to create another version of ourselves.

The Earth People achieve this end by the more pleasurable acts associated with procreation!

Well, I busted up both my knees doing Tai Chi and Kung Fu, so I walk worse than Asimo.

I am considering building a biped, and am fretting over a few things:

The knee, one of the 3 axis 'of the hip, and the ankle seem like they need to be massively strong. I have at my disposal, a bunch of burly coreless motors, which of course would be ideal, but I wanted to publish my plans, and relying on a "one shot deal" style of motor makes this impossible. I also have a supply of some decent automotive motors, that are readily available, but by all appearences, efficiency is really critical, and these things are far from that.

I am looking for the best way to build it, to achieve the end result that I want, not necessarily to emulate nature. I am considering building something like the ED 209 from Robocop. Mind you, I am only taking inspiration from the movie. The squat knee design would require springs to make it worth the while. Using springs to counter the majority of the mass of the beast is just another level of headache, but the squat design gave me an idea of how to provide the necessary amount of freedom, without the compound joints at the hip and ankle.

The idea is to provide a single axis hip, to raise and lower the leg, a single axis knee, and ankle as well. all providing rotation in the same axis. Great you think, a "goosestepping" robot. Well, then add axial rotation between the knee and hip, and between the knee and ankle. This design, would allow shifting it's C/G, and would allow turning as well. It would look like something that was damaged when it walked, but I think it is crazy enough to just work. It would rely on the knee being bent to work.

I have some schematic drawings I just may have to put up on the web.

Mike

Yeah, I think that is an artifact of too few axis in the leg. I think that 7 is a better number. I am just trying to build something "epic", and am flailing over a biped( humanoid or not) or some sort of mega-wakler. I hafe a design roughly fleshed out for a 6x3 walker, made of acrylic, polycarb, and cheap automotive motors, but am also considering an ED209( from Robocop) inspired robot. See my post further below for details.Now if I could only find someone in SoCal who liked to wire things up.

You are probably considering worst case leverage. I don't know about ed209, but in the humanoid you can throw the COG forwards and then back again, while moving from squat to stand, by tilting the torso at the hip. This considerably shortens the leverage on the knees. Then use the hips to straighen yourself up, they don't lift the weight of the legs and are probably the best place to keep the battery.

best regards

Robin G Hewitt

Sounds like you are setting up to replace your own legs in the future.

perspective

I did compile the following leg motions, and the associated muscle groups:

There are three degrees of freedom of the hip. The following muscles contribute: Flexion, the raising of the knee, involves the following: Iliopsoas Tensor fasciae latae Rectus femoris Sartorius Adductor longus Adductor brevis Pectineus

Extension, the straightening of the hip, bringing the knee back down, involves the following: Gluteus maximus Semitendiunosus Semimembranosus Biceps femoris Adductor magnus

Abduction, bringing the foot out sideways, involves the following: Gluteus medius Gluteus minimus Tensor fasciae latae Sartorius

Adduction, bringing the foot back in, involves the following: Adductors Pectineus Gracilis Gluteus maximumus

Transverse abduction, separating the knees while you sit, involves the following: Gluteus maximus Gluteus medius Gluteus minimus Piriformus Obturator externus

Transverse adduction, bringing the knees back together again, involves the following: Adductors Pectineus Gracilis Tensor fasciae latae

Lateral rotation, rotating the leg along it's axis, turing the thigh outward, involves the following: Gemellus superior Gemellus inferior Obturator internus Obturator externus Quadratus femoris Piriformus Gluteus maximus Sartorius Gluteus medius

Medial rotation, rotating the leg along it's axis, turing the thigh inward, involves the following: Tensor fasciae latae Gluteus medius Gluteus minimus

At this point, I have seen too many anatomical cross sections for my own comfort.

Phew! Looks like you've done your homework.:)

IMO studying the anatomy of skeletal and muscular structure is a very good approach to biped leg design and one that i have bin recently looking into as well. But i doubt whether it would be necessary to model every motion control parameter, just as we didnt need to build airplanes with flapping wings to fly. But if we could capture the "essence" of biological biped motion control i think that would be a milestone towards true dynamic bipedal walking robots. On a slightly different note: Recently ive bin reading some books on animal anatomy ( yeah i know, i need to get out more) and found to my surprise that there doesnt appear to be any such thing as a backward knee or what is more commonly called a knee-elbow. The only difference between a forward pointing knee and a so called backward knee is in the length of the leg segments, which of course effects leg geometry. If you look at the skeletal leg anatomy of some animals you will see that the backward pointing knee is actually the ankle ,its just that its bin moved up higher on the leg and the upper thigh bone (ulna?) as bin shortened at the hip. Im under the assumption that this will give the animal an advantage in speed since they are running on their toes and will give better backward thrust. This will prolly also give the animal better shock obsorption but as for stability i dont see any difference.

Im still reading.

This is a classic engineering problem - do you make the prototype resemble the final required design as much as possible (ie self-contained, all electronics, power etc on board), or do you simplify the design as much as possible and see if the basic concept works before weighing it down with other considerations? There's no "right" answer to the problem, it's more a matter of style and personal preference than anything else.

The way the USA approached their first nuclear submarine is a good illustration. One school of thought was to build a prototype system with all the components, the reactor, heat exchangers and turbines, spread out so they would be easy to access, monitor and modify. The other school of thought was that, even if they worked spread out, there was no guarantee that they'd still work once crammed into a submarine hull, so they wanted to build the prototype within a mock-up hull and make the design as realistic as possible. For the record, they chose the latter option. Personally, I would have chosen the former.

Tom

I don't see how anyone discussing anroid design on comp.robotics.misc could possibly, by any stretch of the imagination, be called a Luddite.

Perhaps 'Luddite' has changed meaning recently, like 'Chauvinist' did? Nicholas Chauvin was part of Napleons war machine and had the philosophy that if the State said it, then it had to be true, for how could the State be wrong. He was later ridiculed, but I dont think he was ever misogynistic :o)

best regards

Robin G Hewitt